Proposal of New Learning Model for Effective Use of Human Error Events

－Using Self-Organization Maps in Aircraft Maintenance Works－

Abstract

The purpose of this paper is to provide learning model to improve proactive action or risk management for hazard which occurs due to human errors. The paper first considers most of accidents are caused by stochastic coincidence of various situations and malfunctions. So complex system approaches could be applied instead of root cause analysis methodology. Then, authors propose a learning model using Self-Organization Maps methodology. Proposed model contain human error expediencies occurred in aircraft maintenance works specified such as “Age”, “Day of duty”, “Experience” or so on. The first reason using Self-Organization Maps is this method is considered to be suitable to non-linear system description as is actions or intentions like as human being. The second reason is to have cluster retrieval function which makes possible to extract similar events. The last reason is to use in a visual way. The appreciation of this learning model could indicate us some characteristics of possible risk about human error which is emerging consequence if nobody take measure. So applying this model, most likely characteristics could be suggested and avail the necessary actions previously. This learning model with many near miss accidents or minor affairs enough to contribute prevention of accidents.